heguanhui opened a new issue, #64149:
URL: https://github.com/apache/doris/issues/64149
### Version
trunk (4b7a09ff6fd)
### What's Wrong?
`DataTypeQuantileStateSerDe::write_column_to_pb` and `read_column_from_pb`
use raw memory copy (`get_data_at` / `insert_data`) to serialize/deserialize
`QuantileState` values. However, `QuantileState` contains heap-allocated
members (`shared_ptr<TDigest>`, `vector<double>`), so copying
`sizeof(QuantileState)` bytes of raw memory does not capture the heap data. On
deserialization, `reinterpret_cast`-ing the incomplete bytes back into a
`QuantileState` results in dangling pointers and corrupted data.
Additionally, `write_column_to_pb` does not set
`PGenericType::QUANTILE_STATE` as the type ID in the `PValues` message.
### What You Expected?
`write_column_to_pb` should use `QuantileState::serialize()` to properly
serialize the logical value, and `read_column_from_pb` should use
`QuantileState::deserialize()` to reconstruct it. The type ID should be set to
`PGenericType::QUANTILE_STATE`.
### How to Reproduce?
Run the BE unit test `QuantileStateSerdeTest.writeColumnToPb` on the unfixed
code. The test fails because the round-trip through `write_column_to_pb` /
`read_column_from_pb` produces corrupted data:
```
Expected equality of these values:
column_quantile_state->get_data_at(j)
except_column->get_data_at(j)
```
The raw memory comparison shows mismatched bytes because the heap data is
lost during serialization.
### Are you willing to submit PR?
Yes, I will submit a PR to fix this.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
